Uploaded image for project: 'Kudu'
  1. Kudu
  2. KUDU-644

Asynchronous transaction Abort() behavior can cause CHECK failure

    XMLWordPrintableJSON

Details

    • Bug
    • Status: Resolved
    • Blocker
    • Resolution: Fixed
    • Private Beta
    • None
    • consensus, tablet
    • None

    Description

      JD saw this CHECK failure on the ITBLL cluster:

      I0306 14:16:19.821579 11953 raft_consensus_state.cc:389] T 928c5b52e44844ffbd0f1a19d9fc52d3 P a858d496ac53491aa1f452da9ed81974 [FOLLOWER]: Aborting all transactions after (but not including): 270123. Current State: Replica: a858d496ac53491aa1f452da9ed81974, State: 1, Role: FOLLOWER
      Watermarks: {Received: term: 8 index: 270129 Committed: term: 8 index: 270119}
      I0306 14:16:19.821650 11953 raft_consensus_state.cc:414] T 928c5b52e44844ffbd0f1a19d9fc52d3 P a858d496ac53491aa1f452da9ed81974 [FOLLOWER]: Aborting uncommitted operation due to leader change: 8.270124
      I0306 14:16:19.821673 11953 raft_consensus_state.cc:414] T 928c5b52e44844ffbd0f1a19d9fc52d3 P a858d496ac53491aa1f452da9ed81974 [FOLLOWER]: Aborting uncommitted operation due to leader change: 8.270125
      I0306 14:16:19.821689 11953 raft_consensus_state.cc:414] T 928c5b52e44844ffbd0f1a19d9fc52d3 P a858d496ac53491aa1f452da9ed81974 [FOLLOWER]: Aborting uncommitted operation due to leader change: 8.270126
      I0306 14:16:19.821704 11953 raft_consensus_state.cc:414] T 928c5b52e44844ffbd0f1a19d9fc52d3 P a858d496ac53491aa1f452da9ed81974 [FOLLOWER]: Aborting uncommitted operation due to leader change: 8.270127
      I0306 14:16:19.821719 11953 raft_consensus_state.cc:414] T 928c5b52e44844ffbd0f1a19d9fc52d3 P a858d496ac53491aa1f452da9ed81974 [FOLLOWER]: Aborting uncommitted operation due to leader change: 8.270128
      I0306 14:16:19.821734 11953 raft_consensus_state.cc:414] T 928c5b52e44844ffbd0f1a19d9fc52d3 P a858d496ac53491aa1f452da9ed81974 [FOLLOWER]: Aborting uncommitted operation due to leader change: 8.270129
      F0306 14:16:19.822353 12872 mvcc.cc:451] Check failed: _s.ok() Bad status: Illegal state: There is already a transaction with timestamp: 2852850 in flight.
      

      Analysis to follow in a comment.

      Attachments

        Activity

          People

            tlipcon Todd Lipcon
            tlipcon Todd Lipcon
            Votes:
            0 Vote for this issue
            Watchers:
            2 Start watching this issue

            Dates

              Created:
              Updated:
              Resolved: